Claimempirical · verifiableA factual claim that could be checked directly against observation or primary records.constitution →importance · notableImportance 0.50, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →

TopicsNeural Network Weight Noise Large language models Latent Knowledge in Language Models Alignment Faking

Adding noise to language-model weights can reveal capabilities suppressed by sandbagging.

Not yet assessed, so this page carries no verdict. That is not a finding for or against the claim. Attention goes where its expected value is highest and someone funds it; nothing has funded an assessment of this claim yet, and anyone can.

Decomposition

This claim has not been assessed yet. Attention goes where its expected value is highest and someone funds it; once this claim's assessment is funded and runs, it may well decompose into subclaims.

or start a mandate for this whole area →

Provenance

Where this claim has been said, linked to its canonical form.

None of its 3 sources has been traced to an origin, or to one another.

Tice et al. (2024) showed that injecting Gaussian noise into model weights can reveal hidden capabilities because noise disrupts the underperformance mechanism while core capabilities remain partially intact.

Option-Order Randomisation Reveals a Distributional Position Attractor in Prompted Sandbaggingpreprintextraction 0.70

Cacioli, Jon-Paul · arXiv.org · Apr 29, 2026 · original

Detection landscape; requires weight access.

Furthermore, weight noising during evaluation serves as a promising white-box detection technique, disrupting the capability suppression.

Exploration Hacking: Can LLMs Learn to Resist RL Training?preprintextraction 0.85

Jang, Eyon et al. · arXiv.org · Apr 30, 2026 · original

Gaussian noise added to LoRA adapter weights; triggered (sandbagging) condition shows transient accuracy improvement or distinct collapse versus control.

Furthermore, noise injection (see Figure 5), a common heuristic defense, does not elicit the capability

AI Models Can Provably Hide Arbitrary Capabilitiesdeniesconference paperextraction 0.75

Draguns, Andis et al. · arXiv.org · Oct 9, 2026 · original

Analysis of weight-noising elicitation on compiled models; prior work proposed noise injection to reveal sandbagging.

Cite this claim: a formal citation with its evidence attached

Embed this claim: show it on another site with its current assessment

An embed reports the graph's judgment as it stands when the page is viewed: the status with its confidence, how disputed the claim is, and the date. It links back here.

CardFits its column and sizes itself. Without the script it stays a quotation with its link.

<blockquote class="minerval-claim" data-claim-id="fc451329-02cc-4d5f-a155-d3adb0e7cc46"><a href="https://minerval.ai/claims/fc451329-02cc-4d5f-a155-d3adb0e7cc46">Adding noise to language-model weights can reveal capabilities suppressed by sandbagging.</a></blockquote>
<script async src="https://minerval.ai/embed.js"></script>

IframeWhere scripts are not allowed. A fixed height; add ?theme=light or ?theme=dark to fix its colours.

<iframe src="https://minerval.ai/embed/claims/fc451329-02cc-4d5f-a155-d3adb0e7cc46" title="Minerval: Adding noise to language-model weights can reveal capabilities suppressed by sandbagging." width="480" height="200" style="border:0;max-width:100%" loading="lazy"></iframe>

BadgeAn image for a README or a forum post, in Markdown.

[![Minerval's judgment of this claim](https://minerval.ai/embed/claims/fc451329-02cc-4d5f-a155-d3adb0e7cc46/badge.svg)](https://minerval.ai/claims/fc451329-02cc-4d5f-a155-d3adb0e7cc46)

The badge now:

Contribute

Every judgment on this page is open to challenge. A contribution is evaluated on its merits by the reviewer; if it succeeds the page changes, and if it does not, the reasons are stated. Either way the exchange becomes part of the claim’s public record.

Stated views None yet. A view stated to the graph is shown on this page in its speaker’s words, as a source like any other, and carries no weight in the assessment.

Created by extractor · Nov 13, 2026. Every judgment on this page is accompanied by a reasoning trace.